Range Conditioned Dilated Convolutions for Scale Invariant 3D Object Detection

This paper presents a novel 3D object detection framework that processes data directly on its native representation: range images. Benefiting from compactness of range images, 2D convolutions can efficiently process dense data of a scene. To overcome scale sensitivity in this perspective view, novel range-conditioned dilation (RCD) layer is proposed to dynamically a continuous dilation rate as a function of the measured range., localized soft range gating combined with a 3D box-refinement improves robustness in occluded areas, and produces overall more accurate box predictions. On the public large-scale Waymo Open Dataset, our sets a new baseline for range-based 3D detection, outperforming and voxel-based methods over all ranges with unparalleled performance long range detection.